These audio files accompany Mix Rescue in SOS February 2012.
sos-mix-rescue-feb12-mp3.zip | Original MP3 41 MB sos-mix-rescue-feb12-wav.zip | SOS Remix WAV 259 MBLVMale01_NoProc
Here's the male lead vocal. As you can hear, the quality of the recording and performance are pretty good, despite the odd lumpy edit here and there.
LVMale02_EQ
Because I started the mix with the vocal, and built everything else around that, very little processing was applied to the track during the final mix. As far as EQ goes, there was only some 65Hz high-pass filtering and an octave-wide 2.5dB peaking cut at around 170Hz, both implemented using Cockos ReaEQ. If you compare this with the LVMale01_NoProc file, you'll hear that the difference isn't enormous — mostly just a case of matching the vocal tone a little closer to the commercial tracks that Sambasevam had earmarked as his reference points.
LVMale03_Multicomp
The only remaining concern I had with the male lead vocal's basic sound was that it was a little too variable tonally, as if the singer was moving rather a lot relative to the mic while performing, and I also wanted to try to bring out the internal details of the sound to make the most of these in the final mix. For this reason I processed the track with the Cockos ReaXcomp multi-band compressor, applying 1-2dB of low-threshold 1.2:1 compression across four bands, crossing over at 350Hz, 3.3kHz, and 8.8kHz. Time constants for the four bands were tweaked to suit the characteristics of the signal in each band, so attack times were (low to high) 15ms, 7ms, 3ms, and 2ms respectively, while release times were 101ms, 74ms, 33ms, and 8ms respectively. A couple of extra decibels of output level in the topmost band helped emphasise the 'air' in the sound, which wasn't quite as prominent as with the track's subsequent female lead vocal.
VocalMasking01_Context
Here's a section of the final remix featuring the two main lead vocals. In order to achieve good high-frequency clarity for these parts, I did quite a lot of processing of potentially conflicting backing parts in order to rein in their high-frequency energy (to avoid too much masking of the vocals) and their transient information (to avoid distracting from the vocals). To hear how much these tactics affected the final sound, compare this file with the VocalMasking02_HFEQOut and VocalMasking03_TransProcOut files.
VocalMasking02_HFEQOut
This audio example features the same section of the final remix as in the VocalMasking01_Context file, but with most of the high-frequency backing-instrument EQ cuts bypassed. Notice how the vocals sound considerably less clear and up-front now by comparison, even though I've not changed their processing or effects at all, simply because they are being masked much more by high-frequency energy from the other parts of the arrangement.
VocalMasking03_TransProcOut
For this demonstration, I've again taken the same section of the final remix that you can hear in the VocalMasking01_Context example file, but this time I've bypassed most of the processing I applied for the purpose of mellowing out spikier sonic elements — processes such as transient reduction, fast compression, and high-frequency limiting. In this case, the transients now generate a good deal of unwelcome distraction from the main vocal parts, which I didn't consider very desirable in such a vocal-led song as this.
LVMale04_AutoEQIn
The final stage of lead vocal processing was automation, in this case both fader automation and dynamic adjustment of the gain of a wide high-frequency peaking filter in Cockos ReaEQ. If you comparing this file with LVMale03_Multicomp you can already hear how this evens out the tone and level of the track in solo, but for an in-context comparison check out LVMale05_AutoEQInContext and LVMale06_AutoEQOutContext.
LVMale05_AutoEQInContext
Here's what the fader and EQ automation spotlighted in the LVMale04_AutoEQIn file sounded like in the context of the final remix. Notice how the lead vocal keeps a very solid position in the mix, which wouldn't have been possible without it — as you can hear if you compare with the LVMale06_AutoEQOutContext audio example.
LVMale06_AutoEQOutContext
This audio example demonstrates what the LVMale05_AutoEQInContext file would have sounded like had I bypassed the fader and EQ automation. Compare the two files side by side to hear the subtle but important effects of this, ideally using a single-point mono speaker, which tends to focus on vocal-balance issues most clearly.
Bass01_Unproc
This is Sambasevam's original programmed bass part, generated via an Acoustic Bass patch from Native Instruments Kontakt 4 software sampler. Although I liked the musical aspects of the part a great deal — it's an imaginative part, and feels like a real selling-point for the sound — I did have reservations about its tonal consistency and its ability to translate to small playback systems.
Bass02_LFNarrow
My first processing step for the bass was to narrow the stereo image of the sample below 90Hz or so to scotch a bit of a phase mismatch between the left and right channels — such mismatches can encourage low-end inconsistency on some speaker systems. For this task I chose GVST's simple freeware GStereo plug-in.
Bass03_LFControl
The second stage of processing to even out the rather wayward low-frequency levels of the bass part involved some stiff compression below 110Hz using the low band of the Cockos ReaXcomp multi-band compressor. I chose a ratio of 4:1 and applied around 12dB of gain reduction with a 15ms attack time and a 74ms release time, but as you can hear, the actual effect isn't nearly as drastic as those figures imply, because the plug-in wasn't affecting the midrange of the timbre at all.
Bass04_SmallSpeakers
In order to improve translation to smaller speaker systems, I applied two different plug-ins to generate additional midrange harmonics, because as things stood the bass sound was relying too much on its low harmonics to hold its place in the balance. First I applied a good dose of analogue-style saturation from Variety Of Sound's freeware Nasty VCS plug-in (and also dialled in some general compression there to increase overall evenness and sustain for the part as a whole) and then I reached for Universal Audio's specialist bass enhancer Precision Hz. Compare this example to the Bass03_LFControl file on different-sized speakers to judge the effectiveness of this processing for yourself.
Bass05_DynamicEQ
Although I was pretty happy with the sound in the Bass04_SmallSpeakers example, there was one final processing stage I resorted to in the final stages of the remix to consolidate the bass tone one step further: dynamic EQ from Platinumears' freeware IQ4 plug-in. The main frequency that still seemed to be resonating in a poorly controlled manned was around 155Hz, so I trained an EQ peak on that and set it to compress moderately at a 3:1 ratio, taking of 6dB during peaks. A further subtle 1.1:1-ratio dynamic peak at around 274Hz seemed to help too, but that there wasn't much in it. Overall, though, the effect on the final sound was significant.
LVMale07_FizzverbSolo
In this audio example you can hear the soloed effect return of the 'fizzverb' I used to provide stereo width expansion and high-frequency enhancement to the lead vocal. The raw reverb is an impulse response of the Lexicon 480L hardware reverb's Small Plate algorithm, running in Christian Knufinke's SIR2 convolution engine. Using SIR2's editing options, I've shortened the reverb to less than one tenth of a second in length, and then effectively high-pass filtered it at around 10kHz, and then I've added extra harmonics to its input signal with an instance of the Cockos ReaJS running an Exciter algorithm. To avoid emphasising the vocal sibilance with such a bright effect patch, I've also inserted Digital Fishphones' freeware Spitfish de-esser before the SIR2 instance. To hear this effect mixed with the lead vocal itself, check out the LVMale08_FizzverbIn file
LVMale08_FizzverbIn
This is what the dry lead vocal from the remix sounds like with the 'fizzverb' processing spotlighted in the LVMale07_FizzverbSolo file. Compare this audio example with LVMale09_FizzverbOut to hear the degree of different the fizzverb makes to the 'breathiness' of the vocal timbre.
LVMale09_FizzverbOut
The lead vocal from the LVMale08_FizzverbIn audio example, but with the 'fizzverb' effect bypassed for comparison purposes.
LVMale10_AllFX
Here I've soloed the final male lead vocal, complete with all the effects and automation used in the final remix. Listen in particular to how the reverb and delay levels are being adjusted during different sections of the song.
LVMale11_AllFXContext
This audio illustration shows how the reverb/delay effects you can hear in the LVMale10_AllFX file melt into the full mix, and how the effect-level automation adapts the vocal presentation to suit different arrangement contexts.
OriginalMix
Here's Sambasevam's original mix of 'Kaathaadi', which he sent in to Mix Rescue to ask advice about improving the sonics.
Remix
This is my remix of 'Kaathaadi' using Sambasevam's original multitrack recordings, carried out within Cockos Reaper with the help of a variety of third-party freeware and payware plug-ins.